Journal of Constructional Steel Research 65 (2009) 1909–1920

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

journal homepage: www.elsevier.com/locate/jcsr

Finite element modeling of push-out tests for large stud shear connectors
Huu Thanh Nguyen, Seung Eock Kim ∗
Department of Civil and Environmental Engineering, Sejong University, 98 Kunja-dong, Kwangjin-ku, Seoul, 143-747, South Korea

article info abstract

Article history: An accurate nonlinear finite element model of the push-out specimen has been developed to investigate
Received 12 May 2009 the capacity of large stud shear connectors embedded in a solid slab. The material nonlinearities of
Accepted 15 June 2009 concrete, headed stud, steel beam and rebar were included in the finite element model. The damage
and failure were included in the material model for the headed stud to accurately obtain the ultimate
Keywords: strength of the stud connector. The capacity and ductility of the connection, the load–slip behaviour and
Composite bridge
failure mode of the headed stud were predicted. The results obtained from the finite element analysis were
Finite element modeling
Headed shear stud
verified against experimental results of other researches. An extensive parametric study was conducted
Push-out test to study the effect of the changes in stud diameter and concrete strength on the capacity and behaviour
Shear connection of the shear connection. The capacity and ductility of the shear connection obtained from the finite
analysis were compared with those specified in EC4 and AASHTO LRFD. It is observed that the AASHTO
LRFD specifications overestimated the capacity of the large stud shear connectors, whereas the design
rules specified in Eurocode-4 were generally conservative for stud diameters of 22, 25 and 27 mm, and
unconservative for diameter of 30 mm. The ductility of the large stud shear connectors is sufficient for
practical application in composite bridges.
© 2009 Elsevier Ltd. All rights reserved.

1. Introduction span composite bridges. Badie et al. [8] performed a push-out

test on large shear studs of 31 mm. Li and Krister [9] conducted
Headed shear connectors are the most common type of shear an investigation of large stud shear connections in high strength
connector used in composite beams and bridges. Push-out tests concrete. Topkaya et al. [10] studied the shear stud strength at
are commonly used to determine the capacity of the shear early concrete ages. Shim et al. [11] investigated the behaviour of
connection and load–slip behaviour of the shear connectors. shear connections in composite beam with full-depth precast slab.
Several researchers performed push-out tests to investigate the Shim et al. [12] conducted the design of connection in composite
behaviour of small headed studs in a composite beam with a steel and concrete bridges with precast decks. Shim [13], Lee
solid slab and profile steel sheeting [1–6]. The small studs with et al. [14] performed the push-out tests to study the static and
a diameter smaller than 22 mm are popularly used in composite fatigue behaviour of large stud shear connectors for steel–concrete
bridges. The behaviour of the small studs is well understood and composite bridges.
their capacity can be predicted by using the formulations specified Push-out tests are usually used to investigate the behaviour of
in design rules. The use of large studs greater than 22 mm in shear connections. However, they are often costly and time con-
diameter could provide considerable advantages and conveniences suming. Finite element modeling of shear connection can provide
in composite bridges. In the high shear area, using the higher an efficient alternative to full scale push-out tests. Many researches
capacity shear connectors as large studs would reduce the number investigated the behaviour of stud shear connections using the
of studs and thus reduce welding time. Small number of large finite element model. Ellobody [15], and Lam and Ellobody [16]
developed an accurate nonlinear finite element model to investi-
studs could help to reduce the deterioration of concrete slabs and
gate the behaviour of headed shear stud connectors in solid slabs.
enhance the safety of field workers because of the large space on
The results obtained from the finite element analysis compared
the top flange.
well with the experimental results conducted by these authors.
Few researches conducted on the studs larger than 22 mm
The capacity of the shear connection, the load–slip behaviour of
in diameter have been found. Hanswille [7] conducted several
the headed studs and the failure modes were accurately predicted
push-out tests on 25 mm stud in the study of small and medium
by the finite element model. Ellobody and Lam [17] modeled the
headed stud in a composite beam with precast hollow core slab.
The results obtained from finite element analysis compared well
∗ Corresponding author. Tel.: +82 2 3408 3291; fax: +82 2 3408 3332. with the experimental results conducted by them. Ellobody and
E-mail address: [email protected] (S.E. Kim). Young [18] investigated the performance of shear connection in
0143-974X/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jcsr.2009.06.010
1910 H.T. Nguyen, S.E. Kim / Journal of Constructional Steel Research 65 (2009) 1909–1920

shear connection in composite beams with solid slab. The results

Nomenclature obtained from the finite element analysis were compared with
design strengths calculated using European Code (EC4) [23] and
Asc Cross-sectional area of headed stud shear connector AASHTO LRFD [24] for headed stud shear connectors in composite
d Diameter of headed stud shear connector bridges.
df Displacement at failure of cohesive element
Ec Initial Young’s modulus of concrete
2. Geometry of push-out test
Ecm Mean value of the secant modulus of concrete
tabulated in the EC4
The push-out test specimen in the experiment study of Pil-Goo
Es Initial Young’s modulus of Steel
Lee et al. [14] was investigated in this study. This specimen is in
fck Compressive cylinder strength of concrete
accordance with the standard push-test specimen in Eurocode-
fct Tensile strength of concrete
4 [23]. The geometry of the specimen is shown in Fig. 1. The width
fcu Compressive cube strength of concrete and thickness of the concrete slab is 600 and 200 mm, respectively.
fu Specified minimum ultimate strength of the headed The thickness of the steel beam is 14 mm. The rebar diameter is
stud connector 16 mm. The height of stud is 155 mm. The stud diameter is of 22,
Gcm Shear modulus of concrete 25, 27 and 30 mm.
Knn , Kss , Ktt Elastic constitutive stiffness of cohesive element
P Load per shear connector
3. Finite element model
Pmax Maximum load per shear connector
PAASHTO LRFD Nominal unfactored design strength calculated
3.1. General
using the American specification
PEC 4 Nominal unfactored design strength calculated
In order to obtain accurate results from the finite element
using European Code analysis, all components in the shear connection must be properly
PFE Capacity of shear connection per stud obtained from modeled. In this study, the finite element program ABAQUS
finite element analysis was used to simulate the push-out test. The main components
PTest Capacity of shear connection per stud obtained from influencing the behaviour of the shear connection in the composite
push-out tests beam are concrete the slab, steel beam, rebar and shear connectors.
r Reduction factor The interaction between components is also very important. Both
tn , ts , tt Tractions of the cohesive element geometric and material nonlinearity were included in the finite
Sult Maximum slip at failure element analysis.
X , Y , Z Local coordinates
εc Strain of concrete
3.2. Finite element type and mesh
εc1 Strain associated with compressive cylinder strength
of concrete
Due to the symmetry of the specimens, only a quarter of the
εcu Ultimate Strain of concrete
push-out test arrangement with two headed studs was modeled.
εy Yield strain of structural steel and reinforcement
Fig. 2(a) shows the finite element model of a quarter of the
εu Ultimate Strain of structural steel and reinforce-
specimen and Fig. 2(b) shows the full view of the specimen.
ment
The push-out specimen is composed of six components. They
εys Yield strain of the headed stud shear connector
are the concrete slab, steel beam, headed studs, cohesive layer,
εus Ultimate Strain of the headed stud shear connector
reinforcement bars and rigid base. The components were modeled
εn , εs , εt Strains of the cohesive element as separate parts as presented in Fig. 3. The concrete slab part was
εno , εso , εto Critical strains of the cohesive element meshed with solid element C3D8R available in ABAQUS library.
α Concrete compressive strain factor
This element type is an 8-node brick element with reduced-
β Concrete tensile strain factor integration stiffness. Each node has three translational degrees of
σy Yield stress of structural steel and reinforcement freedom (DOF). This element can be used for nonlinear analysis
σys Yield stress of headed stud shear connector including of contact, large deformation, plasticity and failure. The
σus Specified minimum ultimate tensile strength of the steel beam and headed studs were modeled in the same part and
headed stud shear connector meshed with solid element C3D8R. The cohesive layer was meshed
with the 8-node three-dimensional cohesive element COH3D8.
composite beams with profiled steel sheeting using finite element The rebar part was modeled by the truss element T3D2. The base
models. Kim et al. [19,20] developed linear and nonlinear two- block is a rigid part discretized by rigid element R3D4. In order to
dimensional finite element models and a linear three-dimensional reduce the analysis time, the coarse mesh was applied as an overall
finite element model for the composite beam with profiled steel size. The fine mesh was applied at the region around the interface
sheeting. The effects of profiled steel sheeting and the width of between concrete and studs to achieve the accurate results. In the
concrete slab were investigated. The finite element analyses per- headed stud, the mesh size was also reduced at the joint between
formed by the researchers mentioned above are limited to the stud the stud and steel beam where the stud would fail under shear
shear connectors no larger than 19 mm in diameter. force. The overall mesh size was 25 mm and the smallest size was
The main objective of this paper is to develop an accurate about 3 mm. The finite element mesh of the specimen is presented
and efficient nonlinear three-dimensional finite element model in Fig. 3.
to investigate the behaviour of large stud shear connections. The
finite element program ABAQUS [21] was used in the analysis. 3.3. Interaction and constrain conditions
The results obtained from the finite element analysis were verified
against the test results carried out by Gattesco and Giuriani [22], The parts were assembled and located at proper location to
Loh et al. [5] and Lee et al. [14]. Parametric study was performed to make the specimen model in Fig. 2. Appropriate constrains were
investigate the effects of the changes in diameter of headed shear used to describe the interaction between components. As shown
stud and strength of concrete on the strength and behaviour of in Fig. 4(a), the nodes on the surfaces of concrete slab around the
 H.T. Nguyen, S.E. Kim / Journal of Constructional Steel Research 65 (2009) 1909–1920 1911

Fig. 1. Design of push-out test specimen.

Concrete slab

Steel beam

y
x
z

Base block
(a) A quarter of push-out test specimen. (b) Full view of push-out test specimen.

Fig. 2. Model of push-out test specimen.

studs were tied to the surfaces of the studs by the tie constrain. in Fig. 4(b). The thin cohesive layer of 0.05 mm in thickness
Using this constrain, the relative slip between these two surfaces was located between the steel beam flange and concrete slab
was eliminated. In push-out experiment, the steel beam flange as illustrated in Fig. 4(c). One surface of the cohesive layer was
surface contacting to the concrete slab is usually greased to reduce tied to the concrete slab and the other surface was tied to the
the friction. In the analysis, frictionless contact interaction was steel beam flange. As observed in push-out tests, the headed stud
applied to the steel flange and concrete slab surfaces shown failed in the shearing mode. The pull-out force was very small
1912 H.T. Nguyen, S.E. Kim / Journal of Constructional Steel Research 65 (2009) 1909–1920

Cohesion layer Rebar

Steel beam and studs
Element type Element type T3D2
Element type C3D8R
COH3D8

Base block
Concrete slab Element type R3D4
Element type C3D8R

Fig. 3. Finite element type and mesh.

Stud Steel flange Concrete

Concrete surface surface
surfaces
surfaces

(a) Surfaces in tie constrain between concrete and (b) Surfaces in contact interaction between
studs. steel flange and concrete.

Steel flange Cohesion

surface layer

Concrete
surface
Rebar

(c) Surfaces in tie constrain between steel flange and cohesion layer, (d) Rebar Embedded in concrete slab.
concrete slab and cohesion layer.

Fig. 4. Constrain and interaction surfaces.

 H.T. Nguyen, S.E. Kim / Journal of Constructional Steel Research 65 (2009) 1909–1920 1913

Loading
surface
Symmetric
BC surface 2

Symmetric
BC surface 1

3 1 Reference node
of rigid base
(a) 1-axis symmetric (b) 3-axis symmetric (c) Rigid base boundary
boundary condition boundary condition condition and loading surface

Fig. 5. Boundary condition and loading surfaces.

in comparison with the shear force and pull-out failure did not used to predict the unstable and nonlinear collapse of a structure.
occur. Therefore, the contact interaction and cohesive layer were It is an implicit load control method. In the RIKS method, the
used at the interface between steel flange and concrete slab, load is applied proportionally in several load steps. In each load
whereas the tie constrain was applied to the concrete-to-stud step the equilibrium iteration is performed and the equilibrium
interface. Preliminarily, the contact interaction was used at the path is tracked in the load–displacement space. This method is
concrete-to-stud interface, but it did not give the desired result, often used in static analysis and shows to be a strong method for
and the analysis time increased significantly. Rebars were located nonlinear analysis. However, due to the equilibrium iteration, the
inside the concrete slab as shown in Fig. 4(d). The embedded RIKS method consumes much time and computer resource for a
constrain was applied to the rebar and concrete slab. In this relatively large model. Moreover, the convergence problem is often
constrain, the translational DOF of the nodes on the rebar elements encountered when material damage and failure are included and
were constrained to the interpolated values of the corresponding thus the ultimate load could not be achieved.
DOF of the concrete elements. The slip and debond of the rebar In this study, the dynamic explicit analysis method was
was ignored. Contact interaction was applied at the interface employed. Dynamic explicit analysis is a time control method.
between concrete slab and rigid base. In this interaction the friction It is popularly used for the problems of impact, metal forming,
coefficient was taken as 0.25, which was referenced from the study progressing damage and failure of material, and so on. It shows to
of Ellobody et al. [25]. be a powerful solution scheme for discontinuous medium, contact
interaction and large deformation. It has been applied in many
3.4. Loading and boundary conditions problems such as crack and failure of concrete material [26], metal
sheet forming [27], composite laminate impact [28], etc. In spite
Because of the symmetry of the push-out test arrangement, the of being a dynamic method, dynamic explicit analysis is also used
symmetric boundary condition (BC) was applied to the surfaces at for quasi-static analyses. In the dynamic explicit analysis method,
the symmetric planes of the specimen as shown in Fig. 5. The 1- the global mass and stiffness matrices need not be formed and
axis symmetric BC was applied to surface 1 (Fig. 5(a)), for which the inverted so each increment is relatively inexpensive compared to
translational displacement U1 and rotational displacements R2 and the implicit analysis. The size of the time increment is determined
R3 of all nodes on surface 1 were restrained. The 3-axis symmetric dependent on the mesh size and material properties. The analysis
BC was applied to surface 2 (Fig. 5(b)) so that the translational time can be reduced by using mass scaling or increasing the loading
displacement U3 and the rotational displacements R1 and R2 of all rate. Explicit analysis is very efficient for solving discontinuous
nodes on surface 2 were restrained. The rigid base was assumed
and contact problems; thus it is appropriate for push-out test
to be immovable so all DOF of the reference node of the rigid
simulation. The dynamic explicit method can be used to simulate a
base were restricted. In this analysis displacement control was
push-out test with the same loading rate as in the real experiment.
applied. Loading was downward enforced displacement applied to
However, in order to reduce the analysis time the approach of
the top surface of the steel beam as shown in Fig. 5(c). The applied
increasing loading rate was applied in this study. Different loading
displacement was linearly increased by amplitude function. The
rate has been tried and the most appropriated rate was determined
loading rate can be varied by using different amplitude functions.
as 0.01 mm/s.
The slip was measured as the relative displacement between the
nodes on the steel flange and on the concrete slab near by the stud.
The load was measured as the total reaction acting on the loading 3.6. Concrete material model
surface.
The nonlinear behaviour of the concrete material is presented
3.5. Analysis method by an equivalent uniaxial stress–strain curve of concrete as shown
in Fig. 6. Both compressive and tensile stress is shown in this figure.
The RIKS method was often used to investigate the behaviour For concrete in compression, three parts of the curve have been
of shear connection in push-out test. The RIKS method is generally identified. The first part is initially assumed to be in the elastic
1914 H.T. Nguyen, S.E. Kim / Journal of Constructional Steel Research 65 (2009) 1909–1920

Fig. 7. Stress–strain relationship for structural and reinforcement steel.

Fig. 6. Schematic stress–strain relationship for concrete material.

range to the proportional limit stress. The value of the proportional

limit stress is taken as 0.4(fck ), as presented in EC2 [29], where
fck is the compressive cylinder strength of concrete. fck is equal to
0.8(fcu ) and fcu is compressive cube strength of concrete. The strain
(εc1 ) associate with fck is equal to 0.0022, given by EC2 [29]. The
initial young’s modulus is calculated using the empirical equation
(1) given by EC2 [29]. Poisson’s ratio of concrete is taken as 0.2.

Ecm = 9.5(fck + 8)1/3 (Ecm in GPa; fck in Mpa). (1)

The second part of the curve is the nonlinear parabolic portion
starting from the proportional limit stress 0.4(fck ) and going to the
concrete strength (fck ). This part of the curve can be determined Fig. 8. Stress–strain relationship for shear connectors.
from Eq. (2), given by EC2 [29]
In this material model, yielding parts of the stress–strain curve of
kn − n2
 
σc = fck (2) the concrete are treated separately for tension and compression.
1 + (k − 2)n The Concrete Damage Plasticity model assumes a non-associated
where: potential plastic flow. The Drucker–Prager hyperbolic function
εc is used as the flow potential. The material dilation angle (ψ ),
n= eccentricity (ε ) were taken as 20◦ and 0.1, respectively. The ratio
εc1 of biaxial compressive strength to uniaxial compressive strength
εc1 = 0.0022 (strain of the peak compressive stress fck ) (fb0 /fc0 ) was taken as 1.16.
εc1
k = 1.1Ecm × .
fck 3.7. Structural steel, reinforcement steel material properties
The third part of the stress–strain curve is descending part
from fck to a value of rfck lower than or equal to fck , where r is The stress–strain relationship of structural and reinforcement
the reduction factor referred from the study of Ellobody [25] and steel was modeled by the bi-linear curve as shown in Fig. 7. The
his collaborators. r can vary from 1 to 0.5 corresponding to the curve presents a simple elastic–plastic model. The mechanical
concrete cube strength from 30 to 100 MPa. In this study, r is taken behaviour for both tension and compression is assumed to be
as a constant value of 0.85. The ultimate strain (εcu ) of concrete at similar.
failure associated with rfck is equal to αεc1 . According to EC2 [29]
and BS 8110 [30], εcu = 0.0035 (meaning that α = 1.75). Ellobody 3.8. Headed stud shear connector material properties
et al. [25] took the α as 11 in a confined concrete model. In present
study, to achieve the good agreement between numerical load–slip The shear connector material is of great importance in the
curve and experiment data, various values of α have been tried and push-out test simulation. The material was modeled by a tri-linear
the best match was obtain at εcu = 0.01, i.e. α ≈ 5. stress–strain curve as shown in Fig. 8 [6]. The material behaviour
For concrete in tension, the tensile stress was assumed to is initially elastic followed by strain softening and then yielding.
increase linearly with respect to the strain until the concrete crack. The yield stress (σys ) is determined at εys = 0.2% and the ultimate
After the concrete crack, tensile stress decreases linearly to zero. stress (σus ) achieves at εus = 0.6%.
The value of the strain at zero stress can be taken as 10 times the In order to achieve the exact load–slip relationship, the material
strain at failure as suggested in ABAQUS [21] manual. However, damage and failure options were used in the material model for
the more tension stiffening is applied, the more easily numerical the shear stud connectors. Modeling the failure of material requires
solution is obtained. In order to avoid unstable behaviour in the two specifications: the damage initiation criterion and the damage
response of the material model, the strain at zero stress was taken evolution response. In general, the damage initiation criterion
as 0.005 (i.e. β ≈ 50). specifies a critical equivalent plastic strain where the stiffness of
The Concrete Damage Plasticity model available in ABAQUS [21] the material starts to degrade, and the damage evolution describes
material library was used to model the concrete material in the how the stiffness of the material degrades. ABAQUS [21] offers
push-out test simulation. This material model is suitable for the many choices of damage initiation criteria. The damage initiation
material with different yield strength in tension and compression. criterion for fracture of metal was used in this study. It includes
 H.T. Nguyen, S.E. Kim / Journal of Constructional Steel Research 65 (2009) 1909–1920 1915

Table 1
Measured dimensions of headed stud shear connectors.
Tested by Headed stud dimensions
Shank diameter (mm) Overall height (mm) Head diameter (mm) Head height (mm)

Gattesco and Giuriani 19 125 31 9

Loh et al. 19 125 31 9
Lee et al. 25 155 38 11
27 155 41 12
30 155 44 12

K: elastic constitutive stiffness matrix

ε: nominal strain vector.
The constitutive thickness of the cohesive layer was taken as
1. The stiffness Knn was taken as 0.1Ecm , Kss and Ktt were taken
as 0.1Gcm , where Ecm and Gcm are the elastic modulus and shear
modulus of concrete.
Quadratic nominal strain criterion as shown in Eq. (4) was used
as the damage initiation criterion of the cohesive layer.

εs εt
 2  2  2
hεn i
+ + =1 (4)
εno εso εto
where: εno , εso , εto are the critical nominal normal strain and shear
strains.
Fig. 9. Typical traction–separation response.
Displacement type in conjunction with the linear softening law
was used to describe the damage evolution after the initiation
Table 2 criteria was reached. The displacement type of damage evolution
Measured concrete material properties. requires a maximum displacement df at which the cohesive layer
Tested by Concrete properties completely failed. In this study, different values of the critical
Ecm (GPa) fck (MPa) fct (MPa) strains and the displacement at failure were tried, and the values
Gattesco and Giuriani 30.8a 26 2.9 that gave the best agreement between the analysis and experiment
Loh et al. 21.5 26.2 2.45 load–slip curves were εno = 0.0001, εso = εto = 0.0005 and df
Lee et al. 36.65(39.61)a 49.4(64.5) 3.5(4.86) = 0.8 mm.
a
The elastic modulus determined from Eq. (1). The numbers in the parentheses
are for the push-out test of stud diameter 27 and 30 mm. 4. Verification of finite element model

of ductile and shear criteria. Ductile damage criterion requires Gattesco and Giuriani [22] created a special test model to study
specifying the fracture strain, the stress triaxiality and the strain a single stud shear connector of 19 mm diameters. The dimension
rate. The shear damage criterion requires fracture strain, shear of the stud is shown in Table 1. The concrete material properties are
stress ratio and strain rate. Two types of damage evolution, energy presented in Table 2. The material properties of the headed stud,
and displacement types, in conjunction with two softening laws, structural steel and rebar are presented in Table 3. They performed
linear and exponential, were used to describe the progressive the direct shear test under monotonic loading. Two specimens,
damage of the material appropriately. Once the damage criterion FEA-cohesive and FEA-w/o-cohesive, using the proposed FE model
is reached, the stiffness of the material degrades following the were analyzed in this study. The FEA-cohesive specimen took into
softening law. The element will be removed from the mesh account the cohesion between steel flange and concrete slab while
when the stiffness at all integration points reaches the maximum the FEA-w/o-cohesive specimen did not. The damage parameters
degradation. In this study, the parameters of material damage such for the headed stud materials used in the finite element analysis
as fracture strain, stress triaxiality, shear stress ratio, strain rate, as are shown in Table 4. The load–slip curves obtained from the finite
well as damage evolution type and softening law were calibrated element analyses are compared with the test result in Fig. 10. It can
for the best agreement between analysis and experiment load–slip be seen that the FEA-cohesive curve had a good agreement with the
curves. Gattesco and Giuriani’s result, while the FEA-w/o-cohesive curve
showed a little difference until the slip reaches the maximum
3.9. Cohesive element and material properties displacement at failure (df ) of the cohesive elements. As a result,
the cohesion affected only initial stiffness of the specimen. When
The cohesive element defined in terms of traction–separation the slip reached the maximum displacement at failure of the
was used to mesh the cohesive layer. The traction–separation cohesive elements, the cohesive force vanished. Therefore, the load
model in ABAQUS [21] assumes an initially linear elastic behaviour capacity of the specimen was not significantly changed by initial
followed by initiation and evolution of damage as illustrated in cohesion. Both the experimental and numerical results showed a
Fig. 9. The elastic behaviour of the element is written in terms of similar maximum slip at failure.
elastic constitutive matrix that relates the nominal stress to the Loh et al. [5] studied the effects of partial shear connection in
nominal strain. The uncouple traction type as presented in Eq. (3) the hogging moment regions of composite beam by experimental
was used. approach. In their study, three push-out tests using a conventional
test model as specified in EC4 [23] were investigated. The
tn Knn 0 0 εn
( ) " #( )
dimension of the stud and material properties are given in
t= ts = 0 Kss 0 εs = K ε (3)
Tables 1–3. However, different from the EC4 specimen with both
tt 0 0 Ktt εt
concrete slabs directly placed on the base, one of the slabs in
where t: nominal stress vector their specimens was placed on roller support to obtain a more
1916 H.T. Nguyen, S.E. Kim / Journal of Constructional Steel Research 65 (2009) 1909–1920

Table 3
Measured material properties of headed stud, structural and rebar steel.
Tested by Head stud Structural steel Rebar steel
Es (GPa) σys (MPa) σus (MPa) Es (GPa) σy (MPa) Es (GPa) σy (MPa)
Gattesco and Giuriani 208 350 480 210 320 208 400
Loh et al. 208 350 466 210 320 208 510
Lee et al. 208 353 426 210 320 208 400

Table 4
Damage parameters of headed stud steel materials used in FE analysis.
For verification with the test of Ductile damage Shear damage
Fracture Disp at fail (fracture energy) Softening law Fracture Disp at fail (fracture energy) Softening law
strain strain

Gattesco and Giuriani 0.3 (3000) Linear 0.8 1 mm Exponential

Loh et al. 0.3 (1800) Linear 0.7 (3500) Exponential
Lee et al. 1.78 (5000) Linear 0.9 1 mm Exponential

Fig. 10. Comparison of the FE analysis result with Gattesco and Giuriani’s test
result. Fig. 12. Comparison of the FE analysis result with Lee et al.’s test result.

of the stud and material properties are given in Tables 1–3. The
damage parameters for the headed stud are shown in Table 4.
Fig. 12 shows a good agreement between the capacities of headed
stud shear connectors obtained from the tests of Lee et al. [14] and
the proposed FE analyses. The failure mode of shear connection
observed from FE analyses is shearing failure of shanks as shown
in Fig. 13. It compared well with the test results of Lee et al. [14].
The capacity and the maximum slip at the failure of the shear
connections obtained from the tests and finite element analyses
are presented in Table 5 for comparison. Because three tests
were performed for each stud diameter in the experiment of Lee
et al. [14], the average load capacity was used in the comparison.
Good agreement has been achieved between experimental and
numerical results for all of the push-out tests. A maximum
difference of 5% was observed between the experimental and
Fig. 11. Comparison of the FE analysis result with Loh et al.’s test result.
numerical results for the stud of 30 mm diameter in Lee
et al.’s tests. The mean value of Ptest /PFEA ratio is 0.99 with the
conservative estimate of the mean strength of the connectors. corresponding coefficient of variation (CV) of 0.028. The maximum
Accounting for this factor, in this study, the support condition slip at failure (Sult ), measured at the load dropped to 10% below the
was relaxed by reducing the friction coefficient to 0.15, instead peak load, obtained from finite element analysis compared well
of 0.25 in the contact interaction between the concrete slab and with the experiment results as shown in Table 5. As a result, the
rigid base. The average load–slip curve derived from the test of Loh finite element models successfully predicted the shear connection
et al. [5] and the curve obtained from the proposed FE analysis are capacity as well as load–slip behaviour of the headed shear stud
compared well in Fig. 11. with common and large diameter.
Lee et al. [14] investigated experimentally the static and fatigue
behaviour of large stud connectors for steel–concrete composite 5. Parametric study
bridges. In their study, a total of nine static push-out tests were
conducted on three stud diameters of 25, 27 and 30 mm; three It is shown that the finite element model accurately predicted
tests were made for each diameter. The present study used the the behaviour of the large headed shear stud in composite beams
same push-out arrangement as Lee et al.’s tests. The dimension with solid slab. Therefore, a parametric study was conducted to
 H.T. Nguyen, S.E. Kim / Journal of Constructional Steel Research 65 (2009) 1909–1920 1917

Failed elements was

removed from the mesh

Fig. 13. Stud shank failure mode.

Table 5
Comparison of shear connection capacity obtained from tests and finite element analysis.
Stud diameter (mm) Experiment Finite element analysis Ptest /PFEA Tested by
Ptest (kN) max slip at fail (mm) PFEA (kN) max slip at fail (mm)

19 108.8 9.7 109.3 9.9 0.99 Gattesco and Giuriani

19 101 16 101.57 14.4 0.99 Loh et al.
25 176.4 Avg 6.33
25 176.7 180.1 6.72 176.8 8.8 1.02
25 187.3 7.31
27 208.2 Avg 9.19
27 238.5 211.2 8.36 207.5 8.4 1.02 Lee et al.
27 186.9 8.92
30 222.8 Avg 9.39
30 240 232.3 9.24 243.4 8.8 0.95
30 234 9.46
Mean 0.99
CV 0.028

Fig. 14. Capacity of 22 mm stud shear connection from FE analysis, EC4 and ASSHTO Fig. 15. Capacity of 25 mm stud shear connection from FE analysis, EC4 and ASSHTO
LRFD. LRFD.

study the effects on the capacity of shear connection by changing specimens similar to those of Lee et al.’s test were used. The detail
the diameter of headed studs and the concrete strength. The dimensions of the specimens are shown in Fig. 1. A total of 32 push-
1918 H.T. Nguyen, S.E. Kim / Journal of Constructional Steel Research 65 (2009) 1909–1920

Table 6
Stud diameter and concrete strength of the specimens in parametric study.
Specimen Stud diameter d (mm) Concrete strength fck (MPa)

ST22-C25 22 25
ST22-C30 22 30
ST22-C35 22 35
ST22-C40 22 40
ST22-C45 22 45
ST22-C50 22 50
ST22-C55 22 55
ST22-C60 22 60
ST25-C25 25 25
ST25-C30 25 30
ST25-C35 25 35
ST25-C40 25 40
ST25-C45 25 45
ST25-C50 25 50
ST25-C55 25 55
ST25-C60 25 60
ST27-C25 27 25
Fig. 17. Capacity of 30 mm stud shear connection from FE analysis, EC4 and ASSHTO
ST27-C30 27 30
LRFD.
ST27-C35 27 35
ST27-C40 27 40
ST27-C45 27 45
ST27-C50 27 50
ST27-C55 27 55
ST27-C60 27 60
ST30-C25 30 25
ST30-C30 30 30
ST30-C35 30 35
ST30-C40 30 40
ST30-C45 30 45
ST30-C50 30 50
ST30-C55 30 55
ST30-C60 30 60

Fig. 18. Effect of concrete strength on the load–slip behaviour of the 30 mm headed
stud.

Fig. 16. Capacity of 27 mm stud shear connection from FE analysis, EC4 and ASSHTO
LRFD.

out test specimens, as summarized in Table 6, were investigated in

the parameter study. The specimens had different stud diameters Fig. 19. Maximum slip at failure.
as 22, 25, 27, 30 mm and concrete strengths fck = 25 ∼ 60 MPa.
The material properties of headed stud, structural steel and rebar 6. Comparison of FE analysis with design codes
were taken according to Eurocode-4 [23] as listed in Table 7.
The concrete properties are shown in Table 8. In this Table, the The shear capacity of large stud shear connectors obtained from
elastic modulus of concrete was determined by Eq. (1). The damage the parametric study was compared with the nominal unfactored
parameters of the headed stud materials were taken as those in design strength of shear connectors calculated by the design codes
the verification with Lee et al.’s test given in Table 4. The load per of Eurocode-4 [23] and AASHTO LRFD [24]. Eurocode-4 specifies
stud of 32 push-out specimens obtained from the FE analyses are the design strength of headed stud shear connector in composite
summarized in Table 9. beam as the lesser of Eqs. (5) and (6).
 H.T. Nguyen, S.E. Kim / Journal of Constructional Steel Research 65 (2009) 1909–1920 1919

Table 7
Material properties of headed stud, structural steel and rebar of parametric study.
Head stud Structural steel Rebar steel
Es (GPa) σys (MPa) σus (MPa) Es (GPa) σy (MPa) Es (GPa) σy (MPa)
208 350 450 210 430 208 460

Table 8
Concrete material properties of parametric study.
Properties of concrete Concrete grade
C20 C25 C30 C35 C40 C45 C50 C55 C60

fck (MPa) 20 25 30 35 40 45 50 55 60
fct (MPa) 2.2 2.6 2.9 3.2 3.5 3.8 4.1 4.35 4.61
Ecm (GPa) 28.85 30.74 31.94 33.28 34.53 35.68 36.77 37.8 38.78

Table 9
Comparison of shear connection capacity obtained from FE analysis and current codes of practice.
Specimen PFEA (kN) PEC 4 (kN) PAASHTO-LRFD (kN) PFEA /PEC 4 PFEA /PAASHTO-LRFD

ST22-C25 141.3 122.5 165.9 1.153 0.852

ST22-C30 143.8 136.8 171.1 1.051 0.841
ST22-C35 146.0 136.8 171.1 1.067 0.853
ST22-C40 148.0 136.8 171.1 1.081 0.865
ST22-C45 149.5 136.8 171.1 1.093 0.874
ST22-C50 150.9 136.8 171.1 1.103 0.882
ST22-C55 152.4 136.8 171.1 1.114 0.891
ST22-C60 153.6 136.8 171.1 1.123 0.898
ST25-C25 173.1 158.2 214.2 1.095 0.808
ST25-C30 176.8 176.7 220.9 1.001 0.801
ST25-C35 180.2 176.7 220.9 1.020 0.816
ST25-C40 182.3 176.7 220.9 1.031 0.825
ST25-C45 184.9 176.7 220.9 1.046 0.837
ST25-C50 187.1 176.7 220.9 1.059 0.847
ST25-C55 189.3 176.7 220.9 1.071 0.857
ST25-C60 191.3 176.7 220.9 1.082 0.866
ST27-C25 195.6 184.5 249.9 1.060 0.783
ST27-C30 199.5 206.1 257.6 0.968 0.774
ST27-C35 202.8 206.1 257.6 0.984 0.787
ST27-C40 206.3 206.1 257.6 1.001 0.801
ST27-C45 208.7 206.1 257.6 1.012 0.810
ST27-C50 210.9 206.1 257.6 1.023 0.818
ST27-C55 214.6 206.1 257.6 1.041 0.833
ST27-C60 217.2 206.1 257.6 1.054 0.843
ST30-C25 227.0 227.8 308.5 0.997 0.736
ST30-C30 232.2 254.5 318.1 0.913 0.730
ST30-C35 241.8 254.5 318.1 0.950 0.760
ST30-C40 246.0 254.5 318.1 0.967 0.773
ST30-C45 249.0 254.5 318.1 0.978 0.783
ST30-C50 252.3 254.5 318.1 0.991 0.793
ST30-C55 254.0 254.5 318.1 0.998 0.798
ST30-C60 256.4 254.5 318.1 1.007 0.806
Mean – – – 1.035 0.820
CV – – – 0.052 0.051

PEC 4 = 0.8fu π d2 /4


(5) are shown in Table 9 for comparison. It can be seen that the
p AASHTO LRFD overestimated the design strength of all push-out
PEC 4 = 0.29α (fck Ecm ) (6) specimens investigated in the parametric study by up to 27%. The
EC4 predictions were less conservative for the push-out specimens
where fu is the ultimate strength of steel, fck and Ecm are
with stud diameters of 22 and 25 mm, and slightly conservative for
the cylindrical compressive strength and mean secant (elastic)
the push-out specimens with stud diameter of 27 mm. Whereas,
modulus of concrete, respectively, α = 0.2(h/d + 1) ≤ 1.0, h
the EC4 specifications overestimated the design strength of the
and d are the overall height and diameter of the stud. In AASHTO
stud of 30 mm diameter up to 8.7%, except for some cases of push-
LRFD, the nominal shear resistance of one stud shear connector
out specimens with high concrete strength of 50–60 MPa. The
embedded in concrete deck is taken as Eq. (7).
mean values of PFEA /PEC 4 and PFEA /PAASHTO LRFD ratios are 1.035 and
0.82, respectively, with the corresponding coefficients of variation
p
PAASHTO LRFD = 0.5Asc (fck Ecm ) ≤ Asc fu (7)
(CV) of 0.052 and 0.051, respectively.
where Asc is the cross-sectional area of a stud shear connector. Fig. 18 shows the load per stud versus slip relationship of the
Figs. 14–17 present the relationship between the load per stud push-out specimens with a 30 mm stud for different concrete
and concrete strength obtained from the FE analyses and the design strengths. It can be seen that the capacity of the shear connection
rules specified in the EC4 and AASHTO LRFD for the specimens increased while the maximum slip at failure (Sult ) decreased
with stud diameters of 22–30 mm. The shear connection capacities with the increase of concrete strength. The maximum slip at
1920 H.T. Nguyen, S.E. Kim / Journal of Constructional Steel Research 65 (2009) 1909–1920

failure (Sult ) represents the ductility of the shear connection. It [4] Lloyd RM, Wright HD. Shear connection between composite slabs and steel
is very important for the practical applications of the large stud beams. Journal of Constructional Steel Research 1990;15:255–85.
[5] Loh HY, Uy B, Bradford MA. The effects of partial shear connection in the
in composite bridges. In EN 1994-2 [23], a shear connector with
hogging moment regions of composite beams, Part I—Experimental study.
the diameter about 19 mm is taken as ‘‘ductile’’ if it has Sult ≥ Journal of constructional Steel Research 2004;60:897–919.
6 mm. In BS 5950 [31], there is no data of headed stud capacity [6] Loh HY, Uy B, Bradford MA. The effects of partial shear connection in
presented for concrete cube strength exceeding 40 MPa because the hogging moment regions of composite beams, Part II—Analytical study.
of potentially less ductile behaviour of the shear connector. The Journal of constructional Steel Research 2004;60:921–62.
[7] Hanswille G. Composite bridge design for small and medium spans, New types
results of the parametric study showed that the maximum slip at
of shear connection. ECSC-research report 7210-RP/113. Brüssel; 2002.
failure (Sult ) reduced with the increase of concrete strength as in [8] Badie SS, Tadros MK, Kakish HF, Splittgerber DL, Baishya MC. Large shear studs
Fig. 19. However, all the specimens had a maximum slip at failure for composite action in steel bridge girders. Journal of Bridge Engineering
(Sult ) greater than 6 mm even for high concrete strength up to 2002;7(3):195–203.
fck = 60 MPa. According to Eurocode-4, the ductility of the large [9] An L, Cederwall K. Push-out tests on studs in high strength and normal strength
concrete. Journal of Constructional Steel Research 1996;36(1):15–29.
stud connections is sufficient, and the connections are capable for the
[10] Topkaya C, Yura JA, Williamson EB. Composite shear stud strength at
practical application in composite bridges except for haunches where early concrete ages. Journal of Structural Engineering, ASCE 2004;130(6):
the distance to free surfaces is small. 952–960.
[11] Shim C-S, Kim J-H, Chung C-H, Chang S-P. The behavior of shear connection
in composite beam with full-depth precast slab. Structures and Buildings, The
7. Conclusions
Institution of Civil Engineers 2000;140:101–10.
[12] Shim C-S, Chang S-P, Lee P-G. Design of shear connection in composite
Accurate nonlinear finite element models of push-out specimen steel and concrete bridges with precast decks. Journal of Constructional Steel
have been developed to investigate the capacity of large stud Research 2001;57:203–19.
shear connectors embedded in a solid slab. The models took into [13] Shim C-S. Experiments on limit state design of large stud shear connectors.
KSCE Journal of Civil Engineering 2004;8:313–8.
account the nonlinear material properties of the concrete, steel
[14] Lee P-G, Shim C-S, Chang S-P. Static and fatigue behavior of large stud shear
beam, reinforcement bars and headed stud shear connectors. The connectors for steel–concrete composite bridges. Journal of Constructional
material damage and failure models were included for the headed Steel Research 2005;61:1270–85.
stud shear connectors to accurately obtain the ultimate strength of [15] Ellobody E. Finite element modeling of shear connection for steel–concrete
the studs. The initial cohesion was included to properly determine composite girders. Ph.D. thesis. Leeds: School of Civil Engineering, The
University of Leeds; 2002.
the stiffness of the connections. The capacity and ductility of [16] Lam D, Ellobody E. Behavior of headed stud shear connectors in composite
the connection, the load–slip behaviour and failure mode of the beam. Journal of Structural Engineering, ASCE 2005;131(1):96–107.
headed stud were predicted from the finite element analysis [17] Ellobody E, Lam D. Modeling of headed stud in steel-precast composite beams.
and compared well with the experimental results from other Steel & Composite Structures 2002;2(5):355–78.
researches. [18] Ellobody E, Young B. Performance of shear connection in composite beams
with profiled steel sheeting. Journal of Constructional Steel Research 2006;
An extensive parametric study of 32 push-out specimens with 62:682–94.
different stud diameters and concrete strengths was performed [19] Kim B, Wright HD, Cairns R. The behaviour of through-deck welded shear
using the proposed finite element analyses. The headed stud connectors: An experimental and numerical study. Journal of Constructional
shear connection capacities and ductility obtained from the finite Steel Research 2001;57:1359–80.
analysis were compared with the design rules specified in EC4 [20] Kim B, Wright HD, Cairns R. The behaviour of through-deck welded shear
connectors: A numerical study. In: The first international conference on steel
and AASHTO LRFD. The comparison showed that, the AASHTO and composite structures. 2001. p. 1327–34.
LRFD specifications overestimated the capacity of the headed stud [21] ABAQUS standard user’s manual, Version 6.6. USA: Hibbitt, Karlsson and
shear connectors up to 27%. The design rules specified in EC4 were Sorensen; 2006.
generally conservative for the stud diameters of 22 and 25 mm, [22] Gattesco N, Giuriani E. Experimental study on stud shear connectors subjected
less conservative for the stud diameter of 27 mm. Whereas, the to cyclic loading. Journal of Constructional Steel Research 1996;38(1):1–21.
[23] ENV 1994-2. Eurocode-4: Design of composite steel and concrete structures,
EC4 specifications overestimated the design strength of the stud Part 2: Composite bridges. CEN; 1997.
diameter of 30 mm up to 8.7%, except some cases with a high [24] AASHTO LRFD. Bridge design specifications. 3rd ed. American Association of
concrete strength of 50–60 MPa. The ductility of the large stud State Highway and Transportation Officials; 2004.
shear connections was greater than 6 mm, even for the high [25] Ellobody E, Young B, Lam D. Behaviour of normal and high strength concrete-
filled compact steel tube circular stub columns. Journal of Constructional Steel
concrete cylindrical strength up to 60 MPa.
Research 2006;62:706–15.
[26] William A, John L, Conrad I. Perforation of composite floor. In: 5th European
Acknowledgement LS-DYNA users conference. 2005.
[27] Jung DJ. Study of dynamic explicit analysis in sheet metal forming processes
This work was supported by the Korean Ministry of Education, using faster punch velocity and mass scaling scheme. Journal of Materials
Science and Technology under the Brain Korea 21 Project. Engineering and Performance 1998;7(4).
[28] Nguyen M, Elder D, Bayandor J, Thomson R, Scott M. A review of explicit
finite element software for composite impact analysis. Journal of Composite
References
Materials 2005;39:375–86.
[1] Hawkins NM, Mitchell D. Seismic response of composite shear connections. [29] ENV 1992-1-1. Eurocode-2: Design of concrete structures, Part1: General rules
Journal of Structural Engineering, ASCE 1984;110(9):2120–36. and rules for building. CEN; 1992.
[2] Jayas BS, Hosain MU. Behaviour of headed studs in composite beams: Push-out [30] BSI. BS 8110, Parts 1, 2. Code of practice for design and construction. London:
tests. Canadian Journal of Civil Engineering 1987;15:240–53. British Standards Institution; 1997.
[3] Jayas BS, Hosain MU. Behaviour of headed studs in composite beams: Full-size [31] BSI. BS 5950, Part 3: Section 3.1. Code of practice for design of simple and
tests. Canadian Journal of Civil Engineering 1989;16:712–24. continuous composite beams. London: British Standards Institution; 1990.